[LTP] [PATCH v1 03/10] syscalls/ioctl_loop01: Add LO_FLAGS_AUTOCLEAR and LO_FLAGS_PARTSCAN test

[Yang Xu <xuyang2018.jy@cn.fujitsu.com>]

Hi Cyril

> Hi Cyril
> 
>> Hi!
>>> For LO_FLAGS_AUTOCLEAR flag, we only check autoclear fieldvalue in sys
>>> directory and also get lo_flags by using LOOP_GET_STATUS.
>>>
>>> For LO_FLAGS_PARTSCAN flag, it is the same as LO_FLAGS_AUTOCLEAR flag.
>>> But we also check whether we can scan partition table correctly ie check
>>> whether /dev/loopnp1 and /sys/bloclk/loop0/loop0p1 existed.
>>>
>>> Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
>>> ---
>>> ? runtest/syscalls????????????????????????????? |?? 3 +-
>>> ? testcases/kernel/syscalls/ioctl/.gitignore??? |?? 1 +
>>> ? .../kernel/syscalls/ioctl/ioctl_loop01.c????? | 127 ++++++++++++++++++
>>> ? 3 files changed, 130 insertions(+), 1 deletion(-)
>>> ? create mode 100644 testcases/kernel/syscalls/ioctl/ioctl_loop01.c
>>>
>>> diff --git a/runtest/syscalls b/runtest/syscalls
>>> index 777f0ac6a..7fd67450e 100644
>>> --- a/runtest/syscalls
>>> +++ b/runtest/syscalls
>>> @@ -524,10 +524,11 @@ ioctl03????? ioctl03
>>> ? ioctl04????? ioctl04
>>> ? ioctl05????? ioctl05
>>> ? ioctl06????? ioctl06
>>> -
>>> ? ioctl07????? ioctl07
>>> ? ioctl08????? ioctl08
>>> +ioctl_loop01 ioctl_loop01
>>> +
>>> ? ioctl_ns01 ioctl_ns01
>>> ? ioctl_ns02 ioctl_ns02
>>> ? ioctl_ns03 ioctl_ns03
>>> diff --git a/testcases/kernel/syscalls/ioctl/.gitignore 
>>> b/testcases/kernel/syscalls/ioctl/.gitignore
>>> index 97fbb9681..4cfef2839 100644
>>> --- a/testcases/kernel/syscalls/ioctl/.gitignore
>>> +++ b/testcases/kernel/syscalls/ioctl/.gitignore
>>> @@ -6,6 +6,7 @@
>>> ? /ioctl06
>>> ? /ioctl07
>>> ? /ioctl08
>>> +/ioctl_loop01
>>> ? /ioctl_ns01
>>> ? /ioctl_ns02
>>> ? /ioctl_ns03
>>> diff --git a/testcases/kernel/syscalls/ioctl/ioctl_loop01.c 
>>> b/testcases/kernel/syscalls/ioctl/ioctl_loop01.c
>>> new file mode 100644
>>> index 000000000..e806e0443
>>> --- /dev/null
>>> +++ b/testcases/kernel/syscalls/ioctl/ioctl_loop01.c
>>> @@ -0,0 +1,127 @@
>>> +// SPDX-License-Identifier: GPL-2.0-or-later
>>> +/*
>>> + * Copyright (c) 2020 FUJITSU LIMITED. All rights reserved.
>>> + * Author: Yang Xu <xuyang2018.jy@cn.jujitsu.com>
>>> + *
>>> + * This is a basic ioctl test about loopdevice.
>>> + * It is designed to test LO_FLAGS_AUTOCLEAR and LO_FLAGS_PARTSCAN 
>>> flag.
>>> + *
>>> + * For LO_FLAGS_AUTOCLEAR flag, we only check autoclear fieldvalue 
>>> in sys
>>> + * directory and also get lo_flags by using LOOP_GET_STATUS.
>>> + *
>>> + * For LO_FLAGS_PARTSCAN flag, it is the same as LO_FLAGS_AUTOCLEAR 
>>> flag.
>>> + * But we also check whether we can scan partition table correctly 
>>> ie check
>>> + * whether /dev/loopnp1 and /sys/bloclk/loop0/loop0p1 existed.
>>> + */
>>> +
>>> +#include <stdio.h>
>>> +#include <unistd.h>
>>> +#include <string.h>
>>> +#include "ioctl_loop_support.h"
>>> +#include "lapi/loop.h"
>>> +#include "tst_test.h"
>>> +
>>> +static char dev_path[1024], backing_path[1024], 
>>> backing_file_path[1024];
>>> +static int dev_num, attach_flag, dev_fd;
>>> +/*
>>> + *In drivers/block/loop.c code, set status function doesn't handle
>>> + *LO_FLAGS_READ_ONLY flag and ingore it. Only loop_set_fd with rondonly
>>> + *mode file_fd, lo_flags will include LO_FLAGS_READ_ONLY.
>>> + */
>>> +#define set_flags (LO_FLAGS_AUTOCLEAR | LO_FLAGS_PARTSCAN | 
>>> LO_FLAGS_READ_ONLY | LO_FLAGS_DIRECT_IO)
>>> +#define get_flags (LO_FLAGS_AUTOCLEAR | LO_FLAGS_PARTSCAN)
>>> +
>>> +static char partscan_path[1024], autoclear_path[1024];
>>> +static char loop_partpath[1026], sys_loop_partpath[1026];
>>> +
>>> +static void verify_ioctl_loop(void)
>>> +{
>>> +??? int ret;
>>> +??? struct loop_info loopinfo, loopinfoget;
>>> +
>>> +??? tst_attach_device(dev_path, "test.img");
>>> +??? attach_flag = 1;
>>> +
>>> +??? check_sys_value(partscan_path, 0);
>>> +??? check_sys_value(autoclear_path, 0);
>>> +??? check_sys_string(backing_path, backing_file_path);
>>> +
>>> +??? dev_fd = SAFE_OPEN(dev_path, O_RDWR);
>>> +??? memset(&loopinfo, 0, sizeof(loopinfo));
>>> +??? memset(&loopinfo, 0, sizeof(loopinfoget));
>>> +
>>> +??? loopinfo.lo_flags = set_flags;
>>> +??? SAFE_IOCTL(dev_fd, LOOP_SET_STATUS, &loopinfo);
>>
>> I guess that it would be safer to memset() the loopinfo_get here, also
>> if we zero the structure we may as well reuse the loopinfo here as well.
>>
> of coure, I will add memset in here.
> 
>>> +??? SAFE_IOCTL(dev_fd, LOOP_GET_STATUS, &loopinfoget);
>>> +
>>> +??? if (loopinfoget.lo_flags & ~get_flags)
>>> +??????? tst_res(TFAIL, "expect %d but got %d", get_flags, 
>>> loopinfoget.lo_flags);
>>> +??? else
>>> +??????? tst_res(TPASS, "get expected lo_flag %d", 
>>> loopinfoget.lo_flags);
>>> +
>>> +??? ret = access(loop_partpath, F_OK);
>>> +??? if (ret == 0)
>>> +??????? tst_res(TPASS, "access %s succeeds", loop_partpath);
>>> +??? else
>>> +??????? tst_res(TFAIL, "access %s fails", loop_partpath);
>>> +
>>> +??? ret = access(sys_loop_partpath, F_OK);
>>> +??? if (ret == 0)
>>> +??????? tst_res(TPASS, "access %s succeeds", sys_loop_partpath);
>>> +??? else
>>> +??????? tst_res(TFAIL, "access %s fails", sys_loop_partpath);
>>> +
>>> +??? check_sys_value(partscan_path, 1);
>>> +??? check_sys_value(autoclear_path, 1);
>>> +
>>> +??? SAFE_CLOSE(dev_fd);
>>> +??? tst_detach_device(dev_path);
>>> +??? attach_flag = 0;
>>> +}
>>> +
>>> +static void setup(void)
>>> +{
>>> +??? const char *const cmd_dd[] = {"dd", "if=/dev/zero", 
>>> "of=test.img", "bs=1M", "count=10", NULL};
>>> +??? const char *const cmd_parted[] = {"parted", "-s", "test.img", 
>>> "mklabel", "msdos", "mkpart",
>>> +??????????????????????? "primary", "ext4", "1M", "10M", NULL};
>>> +
>>> +??? dev_num = tst_find_free_loopdev(dev_path, sizeof(dev_path));
>>> +??? if (dev_num < 0)
>>> +??????? tst_brk(TBROK, "Failed to find free loop device");
>>> +
>>> +??? SAFE_CMD(cmd_dd, NULL, NULL);
>>
>> Why not tst_fill_file("test.img", 0, 1024 * 1024, 10); ?
>>
> I will try this.

> 
>>> +??? SAFE_CMD(cmd_parted, NULL, NULL);
>>
>> I wonder if we can avoid dependency on parted by having a look at the
>> changes in the image after the call to parted and writing a few bytes
>> ourselves instead.
> I will see kernel code to figure out this, if no other good way, we can 
> split this check into a new single test. it can make this case less 
> couping.
Sorry for misunderstanding,.  I prefer to add a parted test.img(10K is 
ok on my environment) into ltp or add a flag to skip this 
check(/dev/loop0p1) when parted command is not available.  What do you 
think about the two ways?

Best Regards
Yang Xu
>>
>>> +??? sprintf(partscan_path, "/sys/block/loop%d/loop/partscan", dev_num);
>>> +??? sprintf(autoclear_path, "/sys/block/loop%d/loop/autoclear", 
>>> dev_num);
>>> +??? sprintf(backing_path, "/sys/block/loop%d/loop/backing_file", 
>>> dev_num);
>>> +??? sprintf(sys_loop_partpath, "/sys/block/loop%d/loop%dp1", 
>>> dev_num, dev_num);
>>> +??? sprintf(backing_file_path, "%s/test.img", tst_get_tmpdir());
>>> +??? sprintf(loop_partpath, "%sp1", dev_path);
>>> +}
>>> +
>>> +static void cleanup(void)
>>> +{
>>> +??? if (dev_fd > 0)
>>> +??????? SAFE_CLOSE(dev_fd);
>>> +??? if (attach_flag)
>>> +??????? tst_detach_device(dev_path);
>>> +}
>>> +
>>> +static struct tst_test test = {
>>> +??? .setup = setup,
>>> +??? .cleanup = cleanup,
>>> +??? .test_all = verify_ioctl_loop,
>>> +??? .needs_root = 1,
>>> +??? .needs_cmds = (const char *const []) {
>>> +??????? "dd",
>>> +??????? "parted",
>>> +??????? NULL
>>> +??? },
>>> +??? .needs_drivers = (const char *const []) {
>>> +??????? "loop",
>>> +??????? NULL
>>> +??? },
>>> +??? .needs_tmpdir = 1,
>>> +};
>>> -- 
>>> 2.23.0
>>>
>>>
>>>
>>>
>>> -- 
>>> Mailing list info: https://lists.linux.it/listinfo/ltp
>>
> 
> 
>